NA Digest Sunday, October 8, 1989 Volume 89 : Issue 39

Today's Editor: Cleve Moler

Today's Topics:


From: Iain Duff <>
Date: Mon, 2 Oct 89 11:03:37 CDT
Subject: 1989 Fox Prize Meeting


The Fourth Leslie Fox Prize meeting was held at the University
of Cambridge on Monday 4th September immediately preceding the
Silver Jubilee meeting of the IMA. We are grateful to Mike Powell
for hosting the meeting and to ICI for sponsorship.
I think that I speak for all the 39 attendees at the
Meeting when I say that the standard of presentations by
all candidates was truly exceptional. The quality both of
content and presentation was extremely high and the talks
covered many different areas of numerical analysis, some emphasising
the theoretical and others the practical aspects of the subject.
There were more finalists than in any previous year and the
adjudicating committee of Bill Morton (Chairman), John Mason, and
Nancy Nichols had apparently had great difficulty in selecting
this short-list from an entry of 17 papers from six countries,
such was the overall standard of this year's entry.
The quality was such that they almost overshadowed the
numerical analysis minisymposia at the IMA Silver Jubilee Meeting the
following day. Nearly all the more "senior" speakers at that meeting
used vital minutes from their precious thirty to pay
tribute to the speakers at the Fox Meeting whom they regarded
with some awe and a great deal of respect.

The cast of this outstanding performance were (in order of
appearance) ...

A.Edelman (MIT)
Eigenvalues and condition numbers of random matrices
M.Buhmann (Cambridge)
Multivariable cardinal interpolation with radial basis functions
B.L.R.DeMoor (Stanford)
The restricted singular value decomposition: properties and applications
D.J.Higham (Toronto)
Highly-continuous Runge-Kutta interpolants
R.H.Chan (Hong Kong)
Circulant preconditioners for Hermitian Toeplitz systems
A.M.Stuart (Bath)
Linear instability implies spurious periodic solutions
M.Ainsworth (Durham)
An asymptotically exact a-posteriori error estimator for the
finite element approximation of problems with singular solutions

If encores at mathematics meetings were in order, we might still
have been sitting there until this day. After the final curtain
we retired to tea with the usual side stakes on the outcome,
although the only unanimity from the audience was that with talent
such as was displayed at the meeting the future of numerical
analysis was secure.

Unfortunately, because of a family bereavement, Nancy Nichols
had been unable to come to the meeting. The remaining
adjudicators, after one is sure much agonizing, awarded three
first prizes to Martin Buhmann, Bart De Moor, and Andrew Stuart
and second prizes to Mark Ainsworth, Ray Chan, Alan Edelman, and
Des Higham. Leslie Fox presented the prizes.

It may seem that I eulogize too much but let me say personally
that it was a great privilege to be present at such a fitting
tribute to Leslie Fox on the eve of his presentation of a justly
deserved Honorary Fellowship of the Institute of Mathematics and
its Applications.

At the IMAJNA Editorial Board Meeting, which followed the Prize
meeting, it was agreed that Charlie Elliott would replace Bill
Morton, who was retiring after three hardworking years on the
committee. It is planned to hold the Fifth Leslie Fox Prize
Meeting in Dundee on June 24th 1991, the Monday immediately before the
Biennial Conference in Dundee. Further details will be given in
due course.

Alas the quality of the field has severely deplenished the Prize Fund
to which further contribution would be more than gratefully received.
The Fund is quite independent of any organizations but Catherine
Richards of the IMA has graciously agreed to accept contributions on
behalf of the Fund. Contributions great and small can be sent to
her at Southend or to Nancy Nichols at the University of Reading,
Department of Mathematics, Box 220, Reading RG6 2AX.

Iain Duff
Harwell Laboratory
(From IMANA Newsletter 14(1))


From: David Ratner <ingr!>
Date: 3 Oct 89 13:10:30 GMT
Subject: Cost of NETLIB

Can anyone explain what the overall cost might be for downloading
files from netlib via electronic mail? Some of these files can be over 100,000
characters, and alternative methods for obtaining them (such as by .5" tape)
exist. I would like to evaluate the various alternatives in order to find
the cheapest route.

Thanks in advance.
David Ratner (205)772-1879
Intergraph Corp. Huntsville, Al


From: Steve Lederman <super!>
Date: 4 Oct 89 15:25:24 GMT
Subject: Sparse Eigenvalue Problem

The following eigenvalue problem is of interest and methods of obtaining
a solution are desired:

The matrix is block tri-diagonal where each block is of size 30 to 200
with 100 being the average. The entire matrix will be about 1000x1000
in the initial problem. It is desired to get all of the eigenvalues
and eigenvectors of the whole matrix. Both approximate and "exact"
answers would be of use. The initial testing is hoped to run on a VAX
3500 with 16MB so a solution utilizing the sparse nature would be
nice. The problem may get large (2000+) so an efficient algorithm is
desired. (a larger machine would be used.) Our initial ideas are a
block Lanzos or reduction to tridiagonal form. The former is our best
guess for an efficient method that would be easy to implement (i.e.,
standard packages exist). Does anyone have an opinion of other
methods to solve the problem or and opinion on the best method to
choose? All suggestions welcome to the net or directly to me
( Thanks in advance.

Steve Lederman
Supercomputing Research Center, Bowie, Md.


From: Michael Knorrenschild <>
Date: 4 Oct 89 18:14 -0700
Subject: Change of Address for Michael Knorrenschild

My new address is:

Michael Knorrenschild
Dept. of Mathematics
Simon Fraser University
Burnaby, B.C.
V5A 1S6

Phone: (604) 291-4816



Date: Thu, 5 Oct 89 05:36:37 PDT
Subject: Positions at Liverpool in Parallel Numerical Computing


Over the past few months the Centre for Mathematical Software
Research at the University of Liverpool has been awarded several
substantial grants for research into numerical algorithms for
(local memory) MIMD machines. As a result there are currently
available six posts, at both senior and junior level, for research
staff in the Centre. Interested persons should contact either

Dick Wait - (na.wait)
Len Freeman -
or write to
Prof. Mike Delves,
Centre for Mathematical Software Research,
University of Liverpool,
L69 3BX, United Kingdom.


From: Andy Conn <>
Date: Fri, 6 Oct 1989 8:09:42 PDT
Subject: Position at Waterloo

We are currently looking for a new faculty member, likely at the
junior level in the area of combinatorics or optimization.

I would be happy to provide further information if necessary, or if
a candidate would prefer to write directly to the chairman his address is

Professor Ian Goulden
Department of Combinatorics and Optimization,
Faculty of Mathematics,
University of Waterloo,

Telephone : fax (519) 746 6530
Closing date for applications is December 31, 1989

Andy Conn


Date: Fri, 06 Oct 89 17:29 CDT
Subject: Fortran 8X

Query on Fortran 8X:

I would like to receive information on the status of Fortran 8X.
As I recall, a version of it was put out for comment in mid-1988.
What has happened since then?

Ken Atkinson, Math. Dept., University of Iowa

[Editor's Comments: The Fortran 8X situation is controversial, political,
complicated and, right now, at a crucial decision point. One draft
standard submitted to public comment received much support -- and much
opposition. A revised draft is being considered now, and the deadline
for comments is just a few weeks away. There seems to be a possibility
that the American and European standards organizations may not agree.
There is even controversy over where to get a copy of the proposed
standard. This has all been discussed at great length in the
Unix news group on Fortran. There was a workshop on Fortran 8x
last week at Lake Tahoe. Frankly, I don't follow all the technical
details, or all the positions -- although I feel that I should.
I will ask a couple of experts to summarize the situation for
people like Atkinson and me, who are interested, but don't have the
time or inclincation to join the fray. --Cleve.]


From: David Womble <>
Date: 6 Oct 89 08:49:00 MDT
Subject: Fellowship at Sandia, Albuquerque


Mathematical and Computational Science Department
Sandia National Laboratories

Sandia National Laboratories is seeking outstanding candidates in
analytical and computational mathematics to fill its 1990 Research
Fellowship in Applied Mathematical Sciences. The fellowship is
supported by a special grant from the Applied Mathematical
Sciences Research Program at the U.S. Department of Energy.

The fellowship is intended to provide an exceptional research
opportunity for young scientists. Sandia's Mathematics and
Computational Sciences Department maintains strong programs in
analytical and computational mathematics, physics and engineering,
advanced computational approaches for parallel computers, computer
graphics, and computer architectures and languages. Sandia
provides a unique parallel computing environment, including
both a 512-processor and a 1024-processor NCUBE/ten hypercube, a
Connection Machine 2, and several large Cray supercomputers. The
successful candidate must be a U.S. citizen, must have earned a
Ph.D. degree or the equivalent, and should have a strong
background in numerical computation and an interest in advanced
computing research.

The fellowship appointment is for a period of one year, and
may be renewed for a second year. It includes a highly
competitive salary, moving expenses, and a generous
professional travel allowance. Applications from qualified
candidates, as well as nominations for the fellowship, should
be addressed to Robert H. Banks, Division 3531-86AB, Sandia
National Laboratories, P.O. Box 5800, Albuquerque, NM 87185.
Applications should include a resume, a statement of research
goals, and the names of three references.

Further inquiries can be made by calling (505) 846-1546 or by
sending electronic mail to <RCALLEN@SANDIA.GOV>.

Equal Opportunity Employer M/F/V/H
U.S. Citizenship is Required


From: Rick Vaccaro <>
Date: Sat, 7 Oct 89 12:17:46 EDT
Subject: SVD and Signal Processing Workshop

ANNOUNCING the Second International Workshop on


June 25-27, 1990
University of Rhode Island
Kingston, RI USA

The Second International Workshop on Singular Value Decomposition (SVD)
and Signal Processing is an outgrowth of a previous workshop of the same
name which was held in Les Houches, France in September of 1987.
The goal of this workshop is to bring together active researchers in
several fields related to the SVD. This workshop is sponsored by
the College of Engineering at the University of Rhode Island in cooperation
with the ASSP (signal processing) Society of the IEEE, by EURASIP
(the European signal processing society), and by SIAM (society for
industrial and applied mathematics).


Papers are solicited for technical sessions on the following and
related topics:

* Algorithms for the SVD including parallel
implementation, real-time, and adaptive algorithms.
Complexity, accuracy, and convergence.

* Special purpose signal processing applications such as
array processing, model identification, model reduction,
spectrum analysis, and harmonic retrieval.

* Performance analysis and resolution limits of SVD-based
algorithms. Error analysis.

* Special purpose architectures for computing the SVD.

Other topics related to the SVD and its application are welcome.
Authors are invited to submit three copies of a 4-page extended
summary to R.J. Vaccaro (address below) for review. Authors of
accepted papers will be asked to prepare a version for publication
in a conference proceedings. All regular sessions will be poster
presentations. In addition, the workshop will include a number of
invited plenary talks on topics of general interest.


Richard J. Vaccaro (Chairman)
Dept. of Electrical Engineering
University of Rhode Island
KIngston, RI 02881 USA

G. Faye Boudreaux-Bartels
University of Rhode Island

Ed F. Deprettere
Delft University of Technology

Gene H. Golub
Stanford University

Franklin T. Luk
Cornell University

Don Tufts
University of Rhode Island


* Submission of 4-page extended summary . . . January 29, 1990.

* Notification of Acceptance . . . . . . . March 1, 1990.

* Submission of camera-ready paper . . . . . May 15, 1990.

* Workshop Sessions . . . . . . . . . . June 25-27, 1990.


From: Andrew Tron <phoenix!>
Date: 6 Oct 89 20:21:05 GMT
Subject: Looking for Sylvester's Equation Solver

Does anybody in netland know where one can get one's hands on a program
to solve Sylvester's equation? (i.e solve the system AX + XB = C for
a matrix X, given matrices A, B and C). An algorithm to solve it
appeared in CACM as algorithm 432, but it is not available through
Netlib (as far as I know, Netlib only has the ACM algorithms numbered
492 and above). Please respond by email -- I will summarize the
responses if they are of interest.

Thanks in advance.

Andrew Tron at Princeton University (Internet)
uunet!phoenix!awtron (UUCP)


End of NA Digest